Modelling possible causality in the associations between unemployment, cannabis use, and alcohol misuse

Abstract

Background There has been considerable interest in the extent to which substance use and unemployment may be related, particularly the causal pathways that may be involved in these associations. It has been argued that these associations may reflect social causation, in which unemployment influences substance use, or that they may reflect social selection, in which substance use increases the risk of becoming and remaining unemployed. The present study sought to test these competing explanations. Methods Data from the Christchurch Health and Development Study, featuring a longitudinal birth cohort, were used to model the associations between unemployment and both cannabis and alcohol. Data on patterns of unemployment, involvement with cannabis, and symptoms of alcohol use disorder were examined from ages 18–35 years. The associations between unemployment and both cannabis dependence and alcohol use disorder (AUD) were modelled using conditional fixed-effects regression models, augmented by time-dynamic covariate factors. Results The analyses showed evidence of possible reciprocal causal processes in the association between unemployment and cannabis dependence, in which unemployment of at least three months’ duration significantly (p < 0.0001) increased the risk of cannabis dependence, and cannabis dependence significantly (p < 0.0001) increased the risk of being unemployed. Similar evidence was found for the associations between unemployment and AUD, although these associations were smaller in magnitude. Conclusions The present findings support both social causation and social selection arguments, by indicating that unemployment plays a causal role in substance misuse, and that it is also likely that a reverse causal process whereby substance misuse increases the risk of unemployment.